Flocking Control Of Quadrotors

Creatures' clustering is a common phenomenon in nature,such as the flocking of birds,shoaling of fish,the swarming of insects and herd of land animals.Group movement can effectively help creatures to survive,feed and migrate.Inspired by the flocking of birds,we want to control quadrotors to achieve flocking behavior,so that quadrotors can better collaborate with each other to enhance overall capabilities.This paper proposes distributed controllers to control quadrotors to flock without leaders,which can achieve the convergence,collision avoidance and speed matching.Along the way from easy to difficult,this paper firstly studied the quadrotors'flocking based on the two-dimensional simplified model of the quadrotor.Then after the study with the three-dimensional simplified model,we further studied the speed matching problem of the quadrotors' flocking based on the attitude dynamics using a more general three-dimensional model.Finally,software and hardware experiments in the ROS are given to verify the effectiveness of the proposed controllers.The main contents in this paper can be divided into the following four parts:Firstly,movements in the xoy plane are studied.Maintaining in a fixed height,the six-degree-of-freedom model of the quadrotor can be simplified and transformed into a polar coordinate.By choosing an artificial potential energy function,we use the method of feedback linearization to design a position controller,which can make quadrotors converge from different starting points.Then control the magnitude and direction of velocity based on the multi-agent consensus theory,and finally the flocking of quadrotors is achieved in xoy plane.Secondly,quadrotors' flocking control in three-dimensional space is studied based on the six-degree-of-freedom model established in Euler angles.The dynamic system is divided into a position subsystem and an attitude subsystem,which are connected by an anti-resolver.In the position subsystem,based on the controller structure proposed in the multi-agent research,we designed the controller with artificial potential energy function,and use the Lyapunov function to provide the stability.In the attitude sub-system,we choose the backstepping method to design controllers respectively on the three channels of roll,pitch and yaw to track the control objectives,which are given by the anti-resolver.Numerical simulations are included to verify that with the designed controller,the quadrotors achieved the speed match and collision avoidance while con-verging.Thirdly,directly design the torque controller based on the aircraft dynamics,where the attitude of the six-degree-of-freedom model is described by the rotation matrix.Di-rectly from the dynamics of the quadrotor,we design the torque controller with theneighbors' and its own attitude information,and then achieve the speed synchroniza-tion in flocking by controlling the attitude of the quadrotors.The stability of the system can be explained using the Lyapunov function and the LaSalle invariant set theorem.The results of numerical simulations verified the validity of the proposed controller.Finally,construct the software experimental platform,the hardware-in-the-loop experimental platform,and the outdoor experimental platform under the framework of ROS.In the Gazebo virtual physics world of the software experimental platform,the C++ program is written to achieve the '8' formation,the triangular formation and the line formation,and the code frameworks are given in the paper.In the hardware-in-the-loop experimental platform,based on the interaction between the DJI M100 quadrotor's autopilot online and the quadrotors in the Gazebo virtual physics world,the simulation of the convergence process in the flocking control is implemented,which verifies the effectiveness of the algorithm and the stability of the platform.In the outdoor experi-mental platform,convergence process in flocking control is performed by a quadrotor flying in the real world and two quadrotors flying in the Gazebo virtual world.